How Aurora is finding its own lane on the road to autonomous trucking

Sitting in the cab of the largest vehicle rolling down the interstate south of Dallas, I keep looking out at the smallest things: not just other cars, but the remnants of tires and other bits of road debris. 

A different set of eyes is spotting these objects before I do. But they don’t belong to the driver sitting behind the wheel of the Peterbilt that’s taking me for a ride. They’re not eyes at all—instead, an array of sensors attached to the exterior of the rig. 

The company behind those sensors, a Pittsburgh startup called Aurora, plans to make this self-driving system a commercial reality—starting with a route that runs between Dallas and Houston—by the end of this year. 

And unlike some other firms in the race to make self-driving vehicles happen, Aurora isn’t shooting for something like Tesla’s ambitious but so-far-problematic “full self-driving” capability. 

“The proverbial boiling of the ocean in self-driving can be very challenging,” says Sterling Anderson, co-founder and chief product officer at Aurora. “The ocean is too big to boil all at once, solve a specific problem.”

At the starting line in one lane

Aurora has zeroed in on long-haul trucking for a number of reasons: It’s a far larger market than other autonomous-vehicle possibilities, there’s more money to be made, and a general scarcity of drivers. The American Trucking Association estimates that the industry is short of 78,000 drivers and will need to hire another 1.2 million over the next decade. (Bureau of Labor Statistics data show that the median pay for heavy and tractor-trailer drivers in 2023 was $54,320 a year, equating to $26.12 an hour.)

Plus, trucking has an efficiency limit that can’t be fixed by better driver pay or training: how long any one driver can legally stay on the road. 

“Those trucks are inherently limited by what humans can do,” Anderson says. For example, a run from Long Beach, California, to Dallas would take two to three days with a human driver taking mandatory rest stops but just 24 hours with a truck driving nonstop on a single tank of fuel. 

For that reason, Gartner analyst Mike Ramsey sees Aurora’s focus on long-haul trucking as a sensible strategy. Not only does that simplify the task for a self-driving system, trucking’s high operating costs, crash rates, and driver shortage leave it more suitable for automation than robotaxis. “That makes it a good business,” he says.

Aurora picked Texas for its first market after evaluating what Anderson calls the “commercial landscape, technical landscape and regulatory landscape.” With so much freight moving through the Sunbelt, the state having a climate that’s only “occasionally unreasonable,” and its government having “specifically declared self-driving legal,” Texas checked off all three boxes.

Ramsey notes another advantage with Texas: the mostly flat territory between its largest cities. 

“There’s a bunch of humongous cities not very far apart from each other,” he says. “It’s a good place to do it, because it really simplifies the driving.” 

Seeing farther 

Aurora has spent the last few years developing its hardware and software stack. A key technological component of Aurora’s plan sits above the cab of each of its trucks: its proprietary FirstLight lidar sensor, which uses a continuous stream of laser light that varies in frequency instead of the usual fixed-frequency light pulses to detect road objects farther out. 

“When you’re operating an 80,000 pound truck at 75 miles per hour, you have to be able to see at very long ranges,” Anderson says. So where conventional lidar hardware could see about 800 feet down the road, FirstLight’s vision extends to some 1,300 feet.

In terms of human vision. he says, that yields “about 11 seconds of additional heads-up relative to what a human would have.” 

In the cab on Texas’ I-45, FirstLight’s vision—augmented by sets of medium- and short-range lidar sensors, plus radar and more than a dozen cameras—gets reproduced on a monitor above the dashboard and another behind the front passenger seat. A test engineer seated upfront consults the first screen, calling out each movement initiated by the software, while I gawk at the second display.

As we roll along at a steady 65 mph (except for brief accelerations to 75 to pass other vehicles), the system changes lanes to avoid a chunk of a tire but also executes one change with no foreign objects visible. The engineer’s guess: It saw a shadow. Compared to many highway drivers, the truck was predictable, even polite.

After half an hour, in which the truck executes a turnaround via off- and on-ramps and then navigates down a frontage road back to Aurora’s facility (going 45 mph instead of that road’s posted 55 mph limit), the truck stops at a designated point inside Aurora’s lot. A beep chimes in the cab, a light bar in the dashboard turns from green to blue, and the safety driver takes the wheel for the last few dozen yards. 

Traffic ahead

If Aurora can stick to its end-of-the-year timing for commercial service—reiterated in a July 31 investor presentation—it will beat a handful of competitors to making autonomous trucking happen for paying customers. 

One rival, theMountain View, California-based Kodiak Robotics staged its first driverless delivery in July. That took place in a part of Texas with an even simpler scenario for self-driving: private dirt roads in the Permian Basin. 

“Other than the occasional cow, traffic is limited and speeds average less than 20 miles per hour,” founder and CEO Don Burnette wrote in a post at the time. 

Kodiak also announced in August that it had completed 50,000 miles of autonomous operations with safety drivers on board between Dallas and South Carolina in a test with the trucking firm J.B. Hunt and the tire manufacturer Bridgestone.

Other companies include Plus.ai, which has partnered with Hyundai on a yet-to-be-released fuel-cell-powered, heavy-duty truck; Torc Robotics, which is driving for commercialization of its technology in 2027; and Gatik, which is focusing on shorter-distance autonomous trucking and launched a pilot project without safety drivers for Walmart in 2021. 

Little surprise, then, that Augustin Wegscheider, managing director and partner at Boston Consulting Group, is bullish overall on autonomy’s potential for trucking. 

“There’s a real path on the technology side to make it happen,” he says. “And if legislation gets straightened out, I think there’s lots of positive indicators that this can scale.”

That’s especially true for long-haul trucking, Wegscheider adds: “Where you don’t have those eight- to nine-hour downtimes, there’s such a strong upside.” 

But while he sees sticking to fuel-tank radius as a good starting strategy, he endorses moving to such “alternate powertrains” as hydrogen fuel cells to get past that limit.

Further down the road

In early August, a week after my test drive in Texas, Aurora announced a $483 million fund-raising round. 

Anderson, for his part, professes confidence in the company not running out of financial runway and then taking off: “We’re going to get to both unit economic profitability and gross profitability in the next couple of years,” he says. 

Ramsey, however, warns that Aurora may underestimate its own operating costs as it scales up its system: “You have to build a giant operations center to make these autonomous trucks operate and to maintain them.”

He also cautions that established trucking operators may not be as eager to invest in autonomy as Aurora hopes. “Their biggest challenge is the trucking companies—they’re very hesitant,” he says. 

“They’re not just going to jump in on this.” 

Aurora is banking on production efficiencies getting it there. It’s moving to condense FirstLight into a system-on-a-chip design and has signed up the German mobility manufacturer Continental to mass-produce its hardware stack, with a 2027 due date for both projects.

“That gets a dramatic cost reduction for the system itself,” Anderson says. Asked if a cheap, long-range lidar system could give Aurora a second line of business selling it to vehicle manufacturers, he says: “We’ll cross that bridge when we get the luxury of making them.”

The company seems more set on a different follow-up ambition: taking this technology and putting it into much smaller vehicles that transport only people and packages no bigger than suitcases.

“You will have effectively solved all of the problems that you need to solve for a really compelling robotaxi,” Anderson predicts.

“We don’t have to go there,” he says of Aurora’s ambitions. “But we would see it as a shame, having developed what we see as the world’s safest, most effective and efficient system, not to deploy it everywhere it can make a difference.”

https://www.fastcompany.com/91181504/how-aurora-is-finding-its-own-lane-on-the-road-to-autonomous-trucking?partner=rss&utm_source=rss&utm_medium=feed&utm_campaign=rss+fastcompany&utm_content=rss